libtar: Add an istream_t implementation

The tar_istream_t reads the data from a tar file, having been given
the header, and synthesizes zero bytes for sparse regions.

Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>

5 files changed, 331 insertions, 0 deletions

diff --git a/lib/tar/test/data/CREDITS b/lib/tar/test/data/CREDITS
index 7a2738f..dab8951 100644
--- a/lib/tar/test/data/CREDITS
+++ b/lib/tar/test/data/CREDITS
@@ -33,3 +33,6 @@ The following addtional files have been added:
      Contributed in GitHub issue #64. A tar ball that contains a hard link
      where the 100 byte target field is completely filled without containing
      a null-terminator.
+ - istream/sparse.tar
+     Derived from sparse/gnu.tar and contains some test data for testing the
+     tar istream implementation.
diff --git a/lib/tar/test/data/istream/sparse.tar b/lib/tar/test/data/istream/sparse.tar
new file mode 100644
index 0000000..7f4700d
--- /dev/null
+++ b/lib/tar/test/data/istream/sparse.tar
diff --git a/lib/tar/test/tar_istream.c b/lib/tar/test/tar_istream.c
new file mode 100644
index 0000000..cc31282
--- /dev/null
+++ b/lib/tar/test/tar_istream.c
@@ -0,0 +1,73 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * tar_istream.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+#include "io/file.h"
+#include "tar/tar.h"
+#include "util/test.h"
+
+#ifndef TESTUID
+#define TESTUID 1000
+#endif
+
+#ifndef TESTGID
+#define TESTGID TESTUID
+#endif
+
+#ifndef TESTFNAME
+#define TESTFNAME input.txt
+#endif
+
+#ifndef TESTTS
+#define TESTTS 1542905892
+#endif
+
+static const char *fname = STRVALUE(TESTFNAME);
+
+int main(int argc, char **argv)
+{
+	tar_header_decoded_t hdr;
+	char buffer[100];
+	sqfs_s64 ts;
+	istream_t *fp;
+	istream_t *ti;
+	sqfs_s32 ret;
+	(void)argc; (void)argv;
+
+	fp = istream_open_file(STRVALUE(TESTPATH) "/" STRVALUE(TESTFILE));
+	TEST_NOT_NULL(fp);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_ASSERT(read_header(fp, &hdr) == 0);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_EQUAL_UI(hdr.mode, S_IFREG | 0644);
+	TEST_EQUAL_UI(hdr.uid, TESTUID);
+	TEST_EQUAL_UI(hdr.gid, TESTGID);
+	TEST_EQUAL_UI(hdr.actual_size, 5);
+
+	ts = TESTTS;
+	TEST_EQUAL_UI(hdr.mtime, ts);
+	TEST_STR_EQUAL(hdr.name, fname);
+	TEST_ASSERT(!hdr.unknown_record);
+
+	ti = tar_record_istream_create(fp, &hdr);
+	TEST_NOT_NULL(ti);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 2);
+	TEST_EQUAL_UI(((sqfs_object_t *)ti)->refcount, 1);
+
+	ret = istream_read(ti, buffer, sizeof(buffer));
+	TEST_EQUAL_I(ret, 5);
+	buffer[5] = '\0';
+	TEST_STR_EQUAL(buffer, "test\n");
+
+	ret = istream_read(ti, buffer, sizeof(buffer));
+	TEST_EQUAL_I(ret, 0);
+
+	clear_header(&hdr);
+	sqfs_drop(ti);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	sqfs_drop(fp);
+	return EXIT_SUCCESS;
+}
diff --git a/lib/tar/test/tar_istream2.c b/lib/tar/test/tar_istream2.c
new file mode 100644
index 0000000..a3f27d5
--- /dev/null
+++ b/lib/tar/test/tar_istream2.c
@@ -0,0 +1,155 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * tar_istream2.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+#include "io/file.h"
+#include "tar/tar.h"
+#include "util/test.h"
+
+int main(int argc, char **argv)
+{
+	tar_header_decoded_t hdr;
+	char buffer[100];
+	istream_t *fp;
+	istream_t *ti;
+	(void)argc; (void)argv;
+
+	TEST_ASSERT(chdir(TEST_PATH) == 0);
+
+	fp = istream_open_file("format-acceptance/link_filled.tar");
+	TEST_NOT_NULL(fp);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+
+	/* "deep" directory hierarchy containg 2 files */
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_ASSERT(read_header(fp, &hdr) == 0);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_EQUAL_UI(hdr.mode, S_IFDIR | 0777);
+	TEST_STR_EQUAL(hdr.name, "20_characters_here01/");
+	clear_header(&hdr);
+
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_ASSERT(read_header(fp, &hdr) == 0);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_EQUAL_UI(hdr.mode, S_IFDIR | 0777);
+	TEST_STR_EQUAL(hdr.name, "20_characters_here01/20_characters_here02/");
+	clear_header(&hdr);
+
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_ASSERT(read_header(fp, &hdr) == 0);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_EQUAL_UI(hdr.mode, S_IFDIR | 0777);
+	TEST_STR_EQUAL(hdr.name, "20_characters_here01/20_characters_here02/"
+		       "20_characters_here03/");
+	clear_header(&hdr);
+
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_ASSERT(read_header(fp, &hdr) == 0);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_EQUAL_UI(hdr.mode, S_IFDIR | 0777);
+	TEST_STR_EQUAL(hdr.name, "20_characters_here01/20_characters_here02/"
+		       "20_characters_here03/20_characters_here04/");
+	clear_header(&hdr);
+
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_ASSERT(read_header(fp, &hdr) == 0);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_EQUAL_UI(hdr.mode, S_IFREG | 0777);
+	TEST_STR_EQUAL(hdr.name, "20_characters_here01/20_characters_here02/"
+		       "20_characters_here03/20_characters_here04/"
+		       "errored_file_tst");
+	TEST_EQUAL_UI(hdr.actual_size, 5);
+
+	ti = tar_record_istream_create(fp, &hdr);
+	TEST_NOT_NULL(ti);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 2);
+	TEST_EQUAL_UI(((sqfs_object_t *)ti)->refcount, 1);
+	clear_header(&hdr);
+
+	TEST_ASSERT(istream_read(ti, buffer, sizeof(buffer)) == 5);
+	buffer[5] = '\0';
+	TEST_STR_EQUAL(buffer, "test\n");
+	TEST_ASSERT(istream_read(ti, buffer, sizeof(buffer)) == 0);
+
+	ti = sqfs_drop(ti);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+
+	TEST_ASSERT(skip_padding(fp, 5) == 0);
+
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_ASSERT(read_header(fp, &hdr) == 0);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_EQUAL_UI(hdr.mode, S_IFREG | 0777);
+	TEST_STR_EQUAL(hdr.name, "20_characters_here01/20_characters_here02/"
+		       "20_characters_here03/20_characters_here04/"
+		       "some_test_file");
+	TEST_EQUAL_UI(hdr.actual_size, 5);
+
+	ti = tar_record_istream_create(fp, &hdr);
+	TEST_NOT_NULL(ti);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 2);
+	TEST_EQUAL_UI(((sqfs_object_t *)ti)->refcount, 1);
+	clear_header(&hdr);
+
+	TEST_ASSERT(istream_read(ti, buffer, sizeof(buffer)) == 5);
+	buffer[5] = '\0';
+	TEST_STR_EQUAL(buffer, "test\n");
+	TEST_ASSERT(istream_read(ti, buffer, sizeof(buffer)) == 0);
+
+	ti = sqfs_drop(ti);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+
+	TEST_ASSERT(skip_padding(fp, 5) == 0);
+
+	/* "deep" directory hierarchy containg a hard link */
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_ASSERT(read_header(fp, &hdr) == 0);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_EQUAL_UI(hdr.mode, S_IFDIR | 0777);
+	TEST_STR_EQUAL(hdr.name, "20CharsForLnkTest001/");
+	clear_header(&hdr);
+
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_ASSERT(read_header(fp, &hdr) == 0);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_EQUAL_UI(hdr.mode, S_IFDIR | 0777);
+	TEST_STR_EQUAL(hdr.name, "20CharsForLnkTest001/20CharsForLnkTest002/");
+	clear_header(&hdr);
+
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_ASSERT(read_header(fp, &hdr) == 0);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_EQUAL_UI(hdr.mode, S_IFDIR | 0777);
+	TEST_STR_EQUAL(hdr.name, "20CharsForLnkTest001/20CharsForLnkTest002/"
+		       "20CharsForLnkTest003/");
+	clear_header(&hdr);
+
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_ASSERT(read_header(fp, &hdr) == 0);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_EQUAL_UI(hdr.mode, S_IFDIR | 0777);
+	TEST_STR_EQUAL(hdr.name, "20CharsForLnkTest001/20CharsForLnkTest002/"
+		       "20CharsForLnkTest003/20CharsForLnkTest004/");
+	clear_header(&hdr);
+
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_ASSERT(read_header(fp, &hdr) == 0);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	TEST_STR_EQUAL(hdr.name, "20CharsForLnkTest001/20CharsForLnkTest002/"
+		       "20CharsForLnkTest003/20CharsForLnkTest004/"
+		       "01234567890123456789");
+	TEST_ASSERT(hdr.is_hard_link);
+
+	TEST_STR_EQUAL(hdr.link_target, "20_characters_here01/"
+		       "20_characters_here02/20_characters_here03/"
+		       "20_characters_here04/errored_file_tst");
+	clear_header(&hdr);
+
+	/* end of file */
+	TEST_ASSERT(read_header(fp, &hdr) > 0);
+	sqfs_drop(fp);
+	return EXIT_SUCCESS;
+}
diff --git a/lib/tar/test/tar_istream3.c b/lib/tar/test/tar_istream3.c
new file mode 100644
index 0000000..d287081
--- /dev/null
+++ b/lib/tar/test/tar_istream3.c
@@ -0,0 +1,100 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * tar_istream3.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+#include "io/file.h"
+#include "tar/tar.h"
+#include "util/test.h"
+
+static const struct {
+	uint64_t offset;
+	size_t size;
+	int fill;
+} regions[] = {
+	{       0, 4096, 'A' },
+	{  262144, 4096, 'B' },
+	{  524288, 4096, 'C' },
+	{  786432, 4096, 'D' },
+	{ 1048576, 4096, 'E' },
+	{ 1310720, 4096, 'F' },
+	{ 1572864, 4096, 'G' },
+	{ 1835008, 4096, 'H' },
+};
+
+static int byte_from_offset(uint64_t offset)
+{
+	sqfs_u64 diff;
+	size_t i;
+
+	for (i = 0; i < sizeof(regions) / sizeof(regions[0]); ++i) {
+		if (offset >= regions[i].offset) {
+			diff = (offset - regions[i].offset);
+
+			if (diff < regions[i].size)
+				return regions[i].fill;
+		}
+	}
+
+	return '\0';
+}
+
+int main(int argc, char **argv)
+{
+	unsigned char buffer[941];
+	tar_header_decoded_t hdr;
+	istream_t *fp, *ti;
+	uint64_t offset;
+	sqfs_s32 i, ret;
+	(void)argc; (void)argv;
+
+	fp = istream_open_file(STRVALUE(TESTPATH) "/" STRVALUE(TESTFILE));
+	TEST_NOT_NULL(fp);
+	TEST_ASSERT(read_header(fp, &hdr) == 0);
+	TEST_EQUAL_UI(hdr.mode, S_IFREG | 0644);
+	TEST_EQUAL_UI(hdr.uid, 01750);
+	TEST_EQUAL_UI(hdr.gid, 01750);
+	TEST_EQUAL_UI(hdr.actual_size, 2097152);
+	TEST_EQUAL_UI(hdr.record_size, 32768);
+	TEST_STR_EQUAL(hdr.name, "input.bin");
+	TEST_ASSERT(!hdr.unknown_record);
+
+	ti = tar_record_istream_create(fp, &hdr);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 2);
+	TEST_EQUAL_UI(((sqfs_object_t *)ti)->refcount, 1);
+	clear_header(&hdr);
+
+	offset = 0;
+
+	for (;;) {
+		ret = istream_read(ti, buffer, sizeof(buffer));
+		TEST_ASSERT(ret >= 0);
+
+		if (ret == 0)
+			break;
+
+		for (i = 0; i < ret; ++i) {
+			int ref_byte = byte_from_offset(offset + i);
+
+			if (buffer[i] != ref_byte) {
+				fprintf(stderr, "Byte at offset %llu should "
+					"be 0x%02X, but is 0x%02X\n",
+					(unsigned long long)(offset + i),
+					(unsigned int)ref_byte, buffer[i]);
+				return EXIT_FAILURE;
+			}
+		}
+
+		offset += ret;
+		TEST_ASSERT(offset <= 2097152);
+	}
+
+	TEST_EQUAL_UI(offset, 2097152);
+
+	sqfs_drop(ti);
+	TEST_EQUAL_UI(((sqfs_object_t *)fp)->refcount, 1);
+	sqfs_drop(fp);
+	return EXIT_SUCCESS;
+}